https://github.com/jpodivin/kohocarpet

Semi functional implementation of Kohonens Self organizing maps.
https://github.com/jpodivin/kohocarpet

aesthetics artificial-neural-networks machine-learning

Last synced: about 2 months ago
JSON representation

Semi functional implementation of Kohonens Self organizing maps.

• Host: GitHub
• URL: https://github.com/jpodivin/kohocarpet
• Owner: jpodivin
• License: mit
• Created: 2020-07-05T10:07:11.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2023-04-07T19:02:03.000Z (over 1 year ago)
• Last Synced: 2023-04-07T20:11:01.068Z (over 1 year ago)
• Topics: aesthetics, artificial-neural-networks, machine-learning
• Language: Python
• Homepage:
• Size: 43 KB
• Stars: 0
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# KohoCarpet

Admittedly naive implementation of [Kohonens self organizing map](https://en.wikipedia.org/wiki/Self-organizing_map) algorithm.

Not fit for any real use, but the plots of the networks activation values have certain aesthetic appeal.

If you wonder why, there are several reasons. Among them is topology of the map (torus) and that my modification of the algorithm only picks one example per iteration.

Also the metric for selecting neighborhood is all wrong.